© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 24576
    9. 8-RES_FS24576-JB
    11. 20425
    Prev
    Next
    INDEX
    Explanations

    phrases related to suspicious or criminal activity, as well as descriptions of observing and monitoring behavior

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    87
    +0.19
    0.9%
    481
    +0.10
    0.5%
    113
    +0.09
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    526
    +0.19
    0.01
    113
    +0.10
    0.05
    686
    +0.09
    0.04
    Negative Logits
     Investor
    -0.72
     lender
    -0.70
     predecessor
    -0.67
     successor
    -0.66
    ãĤ´ãĥ³
    -0.64
    etheless
    -0.63
    saf
    -0.63
     investor
    -0.62
     presidency
    -0.61
     appointing
    -0.60
    POSITIVE LOGITS
     desks
    0.94
     throats
    0.94
     torches
    0.89
     carts
    0.89
     strangers
    0.89
    their
    0.88
     mouths
    0.88
     themselves
    0.86
     piles
    0.86
     passers
    0.85
    Activations Density 0.654%

    No Known Activations